Inside the main.ipynb:
1.1 : The section includes the brief summary of a paper published in 2018 by Leland McInnes, John Healy and James Melville. The paper was about UMAP ( Uniform Manifold Approximation and Projection ). The technique for dimensionality reduction. Paper: https://arxiv.org/abs/1802.03426
1.2 : An example which implements UMAP. Using the UMAP library, higher dimensions of a dataset are transformed into lower dimensions. Dataset: https://www.kaggle.com/datasets/fernandolima23/classification-in-asteroseismology
2.1 : This section summarize briefly about the Conformer paper published in 2020. Explains about Convolution augmented Transformers. Basically the combination of CNNs and Transformers. Paper : https://arxiv.org/abs/2005.08100
2.2 : A keyword Spotter is built using a pretrained model and a dataset from Huggingface. Model : https://huggingface.co/facebook/wav2vec2-conformer-rel-pos-large Dataset : https://huggingface.co/datasets/google/speech_commands
3.1 : Now, using an audiofeature dataset UMAP is implemneted to reduce the dimensions of the data without loosing much underlying information.